Dimension reduction and model selection for statistically challenging data. This project aims to develop a deep theoretical understanding of the relationship between various dimension reduction and model selection methods used in statistical model building, and then use this understanding to develop new, improved methods of model building for statistically challenging data. The research will impact on both the theory and practice of statistics, and on substantive fields which collect and analyse ....Dimension reduction and model selection for statistically challenging data. This project aims to develop a deep theoretical understanding of the relationship between various dimension reduction and model selection methods used in statistical model building, and then use this understanding to develop new, improved methods of model building for statistically challenging data. The research will impact on both the theory and practice of statistics, and on substantive fields which collect and analyse these kinds of data. This will provide a significant improvement in the statistical model building in areas such as epidemiology, chemical and ecological sciences. The project is timely because of the increasing collection of large-dimensional, complex, correlated data sets in these and many other fields.Read moreRead less
Prediction, inference and their application to modelling correlated data. This project aims to create new, improved methods for prediction and making inference about predictions for a variety of correlated data types through inventing sophisticated and novel resampling schemes such as the generalised fast bootstrap and repeated partial permutation. The research will impact on both the theory and practice of statistics and on substantive fields which use mixed or compositional models to analyse d ....Prediction, inference and their application to modelling correlated data. This project aims to create new, improved methods for prediction and making inference about predictions for a variety of correlated data types through inventing sophisticated and novel resampling schemes such as the generalised fast bootstrap and repeated partial permutation. The research will impact on both the theory and practice of statistics and on substantive fields which use mixed or compositional models to analyse dependent data. This will be a significant improvement in the assessment and stability of statistical models in areas such as social, ecological and geological sciences.Read moreRead less
High Predictive Performance Models via Semi-Parametric Survival Regression. This project will develop novel statistical models for high prediction performance. When applied to help doctor to treat patients, these models allow the users to include gene or other biomarkers for predicting effectiveness of a treatment. When applied to risk management in finance, these models are capable to include an organization's or individual's ongoing finance status to predict, for example, the probability of or ....High Predictive Performance Models via Semi-Parametric Survival Regression. This project will develop novel statistical models for high prediction performance. When applied to help doctor to treat patients, these models allow the users to include gene or other biomarkers for predicting effectiveness of a treatment. When applied to risk management in finance, these models are capable to include an organization's or individual's ongoing finance status to predict, for example, the probability of or time to loan default. Innovative computational methods will be developed for fitting these models. Compared to traditional prediction method, this approach allows greater flexibility while being superior in terms of statistical accuracy and bias. Extensive analyses of healthcare data from diverse fields will be undertaken.Read moreRead less
Choice experiments to improve predictive power for policy makers. In the current economic climate, Australian governments will benefit from superior choice experiments which will lead to improved prediction of the potential public benefit of proposed policy changes. The choice experiments developed here will have a substantial effect on the development of strategies for the promotion and maintenance of a strong health care system as well as being relevant to the maintenance of a sustainable envi ....Choice experiments to improve predictive power for policy makers. In the current economic climate, Australian governments will benefit from superior choice experiments which will lead to improved prediction of the potential public benefit of proposed policy changes. The choice experiments developed here will have a substantial effect on the development of strategies for the promotion and maintenance of a strong health care system as well as being relevant to the maintenance of a sustainable environment, both designated National Research Priority areas. The innovative research proposed will tap into and build strong links with international research networks, advancing Australia's research reputation and providing a rich environment for the training of research graduates.Read moreRead less
Fast approximate inference methods for flexible regression. This project will gear up statisticians to meet challenges brought about by rapid technological change in the 2010s and beyond. It will produce new methodology and theory for fast and flexible analysis of large and complex data sets.
Technology-Driven and Scalable Regression Methodology, Computing and Theory. Regression is a mainstay of data analysis, statistics, machine learning and data science but is in continual need of enhancement in the face of technological change. Scalability and flexibility for the handling of non-linear signals are fundamental to the practical utility of new regression methodology. Several streams of research aimed at confronting data from specific technologies as well as generic types of data are ....Technology-Driven and Scalable Regression Methodology, Computing and Theory. Regression is a mainstay of data analysis, statistics, machine learning and data science but is in continual need of enhancement in the face of technological change. Scalability and flexibility for the handling of non-linear signals are fundamental to the practical utility of new regression methodology. Several streams of research aimed at confronting data from specific technologies as well as generic types of data are proposed. The project is to be networked with researchers in the United States of America and aims to have Australia-based researchers providing leadership in terms of methodological, theoretical, computational and software development.Read moreRead less
Discovery Early Career Researcher Award - Grant ID: DE200100944
Funder
Australian Research Council
Funding Amount
$427,068.00
Summary
Statistical frameworks for high-parameter imaging cytometry data. The project aims to develop statistical and bioinformatics methodology for characterising the complex interactions between cells in their native environment. Recent advances in imaging cytometry technologies have made it possible to observe the behaviour of multiple cell-types in tissue concurrently. The intended outcome is a suite of statistical methodologies that are crucial for addressing a variety of biological problems with t ....Statistical frameworks for high-parameter imaging cytometry data. The project aims to develop statistical and bioinformatics methodology for characterising the complex interactions between cells in their native environment. Recent advances in imaging cytometry technologies have made it possible to observe the behaviour of multiple cell-types in tissue concurrently. The intended outcome is a suite of statistical methodologies that are crucial for addressing a variety of biological problems with these state-of-the-art technologies. This work will advance knowledge in bioinformatics, statistics and image analysis, providing benefits to scientists studying the fundamental behaviour of cells and underlying disease mechanisms.Read moreRead less
Prognosis based network-type feature extraction for complex biological data. This project aims to develop statistical tools that integrate high-throughput molecular data with biological knowledge to make discoveries in complex diseases. This project uses machine learning methods, statistical models and proteomic platforms to identify relationships among clinico-pathologic and molecular measurements. It will produce tools and insights that are intended to accelerate the process of biologically an ....Prognosis based network-type feature extraction for complex biological data. This project aims to develop statistical tools that integrate high-throughput molecular data with biological knowledge to make discoveries in complex diseases. This project uses machine learning methods, statistical models and proteomic platforms to identify relationships among clinico-pathologic and molecular measurements. It will produce tools and insights that are intended to accelerate the process of biologically and clinically significant discoveries in biomedical research. This project will help Australian researchers in statistics and users of statistics (from fields as diverse as biology, ecology, medicine, finance, agriculture and the social sciences) to make better predictions that are easier to understand.Read moreRead less
A likelihood-based approach to combined surveys inference. This project focuses on the development of statistical theory for efficient integration of information across multiple complex sample surveys. It will develop theory and methodology that will answer complex questions about relationships between important social, economic and health related variables that are presently measured in separate surveys.
Statistical Inference for Probability-Linked Longitudinal Data. The Strategic Roadmap for the Australian Government's National Collaborative Research Infrastructure Strategy states that analysis of linked data, and particularly linked longitudinal data, has the potential to revolutionise Australian public health research. Similar benefits should flow from analysis of linked datasets in other areas, e.g. the Statistical Longitudinal Census Dataset that the Australian Bureau of Statistics intends ....Statistical Inference for Probability-Linked Longitudinal Data. The Strategic Roadmap for the Australian Government's National Collaborative Research Infrastructure Strategy states that analysis of linked data, and particularly linked longitudinal data, has the potential to revolutionise Australian public health research. Similar benefits should flow from analysis of linked datasets in other areas, e.g. the Statistical Longitudinal Census Dataset that the Australian Bureau of Statistics intends to create by linking individual records across censuses. These benefits will be maximised by controlling the impact of linkage error when analysing these datasets. This proposal will develop the statistical theory and related methodology to solve this problem in a statistically efficient manner.Read moreRead less